Understanding unstructured text is a major goal within natural language processing. machine comprehension tests (e.g., MCTest) pose questions based on short text passages to evaluate such understanding. Comprehension systems aim to extract the correct answer from the passage which contains it. Partly because of its limited size, prior work on MCTest has focused mainly on feature-based approaches.
It is with respect to these and other general considerations that the aspects disclosed herein have been made. Also, although relatively specific problems may be discussed, it should be understood that the examples should not be limited to solving the specific problems identified in the background or elsewhere in this disclosure.